The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> s-parameters for simulation
https://designers-guide.org/forum/YaBB.pl?num=1168526769

Message started by vivkr on Jan 11^th, 2007, 6:46am

Title: s-parameters for simulation
Post by vivkr on Jan 11^th, 2007, 6:46am

Hi,

Is it possible to simulate a 2-port network with given s-parameters using Spectre? I tried
looking for a primitive in the analogLib but could not see one where the s-parameters could
be directly specified.

Does it actually make sense to use this kind of an approach if one is trying to see say time-domain
or ac behavior and only s-parameters are known, say in form of data from a text file?

Regards
Vivek

Title: Re: s-parameters for simulation
Post by Ken Kundert on Jan 11^th, 2007, 8:24am

Spectre provides a component called an 'nport' that takes a file of s-parameters. In Artist, I believe they used 1port, 2port, ... instead.

The only issue with using a table of s-parameters in a frequency domain simulation is that your should be careful not to ask the simulator to extrapolate the s-parameters. Thus if you are doing a ac or noise analysis the frequency range of the analysis should be within the frequency range of the s-parameters. If you are performing a harmonic balance analysis then the s-parameters should be specified all the way down to DC and up to the highest frequency used by the harmonic balance analysis.

Using tables of s-parameters is problematic with time domain analyses. It can be done if the s-parameters are well behaved but it can be a challenge to get good results.

-Ken

Title: Re: s-parameters for simulation
Post by ACWWong on Jan 11^th, 2007, 8:47am

The spectreRF.pdf in your cadence installation path doc folder has an a chapter dedicated to the use of nport and using tabulated s-parameter data. It goes into some detail with regards the pitfalls of attempting time-domain simulations with tabulated s-paramters.

Title: Re: s-parameters for simulation
Post by Andrew Beckett on Jan 15^th, 2007, 7:11am

There's also an application note on http://sourcelink.cadence.com which can be found by searching for "nport application note" which is a more practical usage guide.

Regards,

Andrew.

Title: Re: s-parameters for simulation
Post by mg777 on Jan 17^th, 2007, 7:53am

S-parameter based models are susceptible to precision errors. For example, passive circuits need to have a unitary S matrix, but if your data comes from a VNA there is no guarantee that S will be 'unitary enough' for (I-S)^-1 to be well conditioned.

You'll be better off fitting a physically meaningful equivalent circuit to the S data, which will also ensure the proper low frequency behavior.

M.G.Rajan
www.eecalc.com